The reliability of empirical conclusions depends heavily on the nature of the work: results based on simple, transparent descriptive data, averages, and reality checks are more trustworthy than those requiring sophisticated statistical techniques with many simplifying assumptions.